UC San Diego

Transcriptome architecture, shaped by mRNA isoforms generated through alternative splicing (AS) and polyadenylation (APA), defines the proteome governing cell states, identities, and responses. Translation is at the center of this relationship, but methodological constraints limit translation profiling at mRNA isoform resolution. To address this, I present Long-read Ribo-STAMP (LR-Ribo-STAMP), a novel approach that combines Ribo-STAMP and long-read sequencing. I develop a computational pipeline for detecting Cytosine-to-Uracil edits in long-read RNA-seq data, optimize methods for improved signal-to-background, and introduce a metric (EditsC) to quantify mRNA isoform-level translation. Using LR-Ribo-STAMP, I globally profile mRNA translation in normoxia and hypoxia in a triple-negative breast cancer cell line, uncovering a hypoxia-induced differential isoform usage resulting in a truncated protein isoform of G-coupled receptor kinase 6 (GRK6). To further elucidate transcriptome regulation, I characterize RNA binding proteins (RBP) in the context of APA, a process that impacts mRNA stability, localization, and translation. Using a high-throughput tethered function assay, I determine RBPs that activate poly(A) site (PAS) selection in an unbiased, location-dependent manner. By fine-tuning the transformer model, ProteinBERT, I provide a framework for using the screen results to predict activators of PAS selection and use occlusion maps to resolve consequential regions of high confidence. Functional and protein-protein interaction analysis reveal two archetypes of high-confidence RBP activators—those directly associated with RNA processing and those with broader functional roles, both with potential unannotated roles in APA. Mechanistic studies on GRB2 and RNPS1, representing these archetypes, further resolve modes through which they may activate PAS selection. Collectively, this work advances methodologies to study translation at the mRNA isoform level, and the characterization of proteins involved in shaping transcriptome architectures, providing opportunities to further our understanding of cellular complexity and function.